Abstract

This work analyzes collaboration patterns and their relationship to research impact in Brazilian computer science, a field that has experienced remarkable growth yet continues to face challenges in achieving global recognition proportionate to its publication volume. Using network science and bibliometric methods over a dataset of 68,142 publications and 128,847 authors collected from OpenAlex, we address three research questions: to what extent do Brazilian computer scientists engage in international collaborations; how do collaboration levels vary among computer science subfields; and how do the structural characteristics of collaboration networks differ across these subfields. Despite high publication output, Brazil's limited collaboration patterns may explain its low citation impact. Our findings reveal that theoretical areas favor international partnerships while applied areas prefer national collaborations. We also identified a few "bridge" researchers serving as key structural elements in high-impact networks. The study maps these networks and demonstrates how strategic international partnerships correlate with research impact, providing insights for improving collaboration and evaluation policies.